Genetic models of Cushing’s disease

Abstract

Cushing’s disease (CD) is caused by pituitary tumors that overproduce adrenocorticotropic hormone (ACTH); however, effective medical treatments remain limited, significantly impairing patients’ quality of life and prognosis. Despite extensive molecular analyses, the pathogenesis of CD remains unclear. Although previous molecular studies have relied heavily on rodent-derived cells and rodent transgenic models, significant species differences exist in the tumorigenesis of CD between humans and rodents. To date, an established human CD cell model is lacking, as human CD cells are limited in availability and sustainability over time. Additionally, the gene modifications used in transgenic models do not necessarily reflect the causative genes in CD. CD tumors exhibit wide phenotypic heterogeneity, which further complicates the development of an ideal genetic model. In this review, we provide an analysis of 11 genetic models used to study CD, outlining their historical development, strengths, and limitations. Additionally, we discuss the ongoing development of human induced pluripotent stem cell (iPSC)-derived pituitary organoids and further describe various models of pituitary organoids as an emerging novel approach to studying CD. By comparing all these models, we highlight the necessity of advancing genetic models to improve our understanding and treatment of CD.

Introduction

Cushing’s disease (CD) is a rare disorder caused by an adenoma of corticotrophs, leading to excessive secretion of adrenocorticotropin hormone (ACTH) from the pituitary gland. Diagnosing and treating CD is challenging. Despite significant molecular analysis, the pathogenesis of CD remains unknown due to the lack of a viable human cell model, and previous molecular studies have consequently relied heavily on non-human models. A deeper understanding of CD pathophysiology can lay the foundation for advancing novel therapeutic approaches to CD and developing human CD models is critical for that understanding. This review examines the physiology of corticotrophs, the genetic basis of CD, and various genetic models, including natural models, transgenic animals, and emerging human cell models using pluripotent stem cells. By highlighting the strengths and limitations of these genetic models, we aim to emphasize the importance of developing new genetic models of CD.

Corticotrophs and the POMC gene in Cushing’s disease

The pituitary gland consists of three lobes: the anterior, intermediate, and posterior [1, 2]. Unlike other species, the intermediate lobe in humans is involuted and non-secretory (Figure 1). CD is characterized by corticotroph adenomas which exhibit elevated POMC expression and a loss of feedback inhibition, leading to excessive ACTH secretion and hypercortisolemia [3]. Current medical therapies primarily manage symptoms by targeting cortisol synthesis or its receptors, with a variable efficacy of 30-80% normalization of urinary free cortisol (UFC) levels. Pituitary adenoma-directed therapies are limited by low potency (e.g., cabergoline, 20-30% normalization of UFC) or unfavorable adverse effects (e.g., pasireotide, hyperglycemia in 70% of cases) [4].

Figure 1. Schematic cross-sectional representation of human pituitary gland and murine pituitary gland.

Images of the cross sectioned human pituitary gland (A) and mouse pituitary gland (B). A: anterior lobe, I: intermediate lobe, P: posterior lobe. Degenerated intermediate lobe in human pituitary is shown by asterisk (*). Created with BioRender.com

Regulation of POMC expression in corticotroph adenomas has been a focal point of research. Key regulating transcription factors include Pituitary Homeobox 1 (PITX1) [12], T-box Transcription Factor TBX19 (TPIT) [5], Neurogenic Differentiation 1 (NEUROD1) [6], and Signal Transducer and Activator of Transcription 3 (STAT3) [7]. Additionally, nuclear receptors such as the glucocorticoid receptor (GR) [8], nuclear receptor subfamily 4A1 (NUR77) [9], orphan nuclear receptor TR4 [10], and E2F1 [11], regulate POMC expression. In addition to regulation by transcription factors, DNA methylation plays a role in POMC transcription [12,13]. Both the conventional POMC promoter (−480/+63) and a recently reported second POMC promoter at the intron 2/exon 3 junction possess CpG island [14]. Hypomethylation of the second POMC promoter correlates with larger adenoma size and a higher recurrence rate [14]. Of note, the majority of these molecular analysis and discoveries were done using mouse ACTH-secreting pituitary adenoma cell line.

Unlike many cancers, CD is typically sporadic, with mutations in well-characterized oncogenes being rare. However, recent studies have reported a strikingly high frequency of gain-of-function mutations in the ubiquitin-specific protease 8 (USP8) gene, occurring in 30-50% of cases [15,16]. The USP family of enzymes regulates post-translational protein modification by protecting substrates from degradation through deubiquitination. Notably, a complete loss of USP8 function is embryonically lethal, underscoring its essential role in cellular functions [17]. However, the deubiquitination targets of USP8 in CD remain not fully elucidated. It has been suggested that the gain-of-function mutations in USP8 may prevent epidermal growth factor receptor (EGFR) from lysosomal degradation, thereby contributing to CD pathogenesis [18], however, clinical and laboratory findings have not always aligned. For instance, EGFR expression levels in USP8-mutated CD are inconsistent, with some exhibiting lower EGFR levels than wild-type (WT) corticotroph adenomas [15]. In addition, many USP8-mutated tumors lack the aggressive phenotype typically associated with EGFR-driven tumors [15,16, 19]. These discrepancies suggest the presence of alternative USP8 substrates in CD that possibly remain unidentified.

Genetic Models of CD.

Corticotroph adenomas are typically small and benign, and to date, human corticotroph adenoma cell lines have not been established. To better understand the mechanisms of corticotroph tumorigenesis and develop effective treatment strategies, genetic models are essential. In this entire section, we introduce 3 non-human natural models, then 6 transgenic in-vivo models, followed by emerging 2 types of human cell-derived models (Table 1, Figure 2).

Table 1.

Genetic models of Cushing’s disease.

Natural models	Canine model
	AtT20 mouse cell line
	DMS79 human cell line (non-pituitary origin)
Transgenic models	CRH mouse model
	Large T antigen mouse model
	Rb knockout mouse model
	pttg zebrafish model
	Egfr mouse model
Human models	Sustainable human tumor primary cell culture
	iPSC-derived human pituitary organoid

Figure 2. Currently available Cushing’s disease genetic models.

Schematic of Cushing’s disease genetic models and their pros and cons. Created with BioRender.com.

Natural models of CD from non-human origins and human non-pituitary cells

Canine Model

Canines exhibit a notably high frequency of CD development, with an estimated annual incidence of 1-2 cases per 1,000 dogs, which is 100 times higher than in humans [20–23]. The underlying mechanisms behind this elevated prevalence remain unknown. Like in humans, CD in canines shows a female predominance. Canine CD symptoms include polyuria, polydipsia, central adiposity, fragile skin, and hair loss [20–22]. Sanders et al. reported that among 3,532 cases of canine pituitary adenomas, 95% (3,387 cases) were corticotroph/melanotroph adenomas [21]. This remarkably high proportion of corticotroph adenomas among pituitary adenomas appears unique to canines, as corticotroph/melanotroph adenomas were found in only 23% of cases in cats in the same study [21]. Further investigation revealed that both corticotrophs in the anterior lobe and melanotrophs in the intermediate lobe seem to contribute to the formation of ACTH-producing adenomas in canines (Figure 1). This cellular origin pattern differs from CD in humans, where the intermediate lobe typically undergoes degeneration and its involvement in adenoma formation has not been reported.

Regardless of their origins (corticotrophs or melanotrophs), adenomas resected from canines have been utilized as models for human CD [24, 25]. Wang et al. studied both canine and human corticotroph adenomas using whole exome sequencing and RNA sequencing, which revealed very similar gene expression profiles between canine and human CDs [25]. Fukuoka et al. utilized primary adenoma cell cultures from canines and found that treating cells with Gefitinib, an EGFR tyrosine kinase inhibitor, suppressed both POMC expression and ACTH secretion [24]. Similar results were observed in primary cell cultures from human CD. Thus far, canines are the only natural in vivo model of CD. While canine adenomas serve as a useful model for human CD, it is essential to acknowledge the difference in cellular origins (corticotrophs or melanotrophs), which may influence the interpretation of data derived from canine CD-featured adenoma studies.

Mouse ACTH-secreting adenoma cell line (AtT-20)

The AtT-20 cell line, a mouse ACTH-secreting pituitary adenoma cell line, was established in the 1950s [26–27], originated from the pituitary of whole-body irradiated mice. AtT-20 cells remained the only in vitro model until the recent development of human pituitary organoids [5–10, 28,29], and it has been widely used to investigate molecular mechanisms regulating Pomc expression. Over the decades, research using AtT-20 cells identified multiple upstream regulators, including Pitx1 [12], Tbx19 (Tpit) [5], NeuroD1 [6], Stat3 [7], Nur77 [9], Tr4 [10], Brg1 [28], and Bmp4-Smad1 [29]. Additionally, AtT-20 cells have been instrumental in screening potential CD therapeutics, both in vitro and in vivo xenograft experiments [24, 30–32].

While the AtT-20 cell line has significantly contributed to the study of CD and Pomc regulation, some studies suggest differences between POMC/Pomc regulation in human corticotroph cells and AtT-20 cells. As discussed in the next section, research on the human DMS79 cell line, which expresses POMC and secretes ACTH, indicates potential regulatory differences. Whether these discrepancies stem from species-specific differences or their distinct cellular origins (AtT-20 cells from corticotrophs, vs. DMS79 cells from lung cancer) remains unclear.

Human ACTH-secreting small cell lung cancer cell line (DMS79)

Although there are no POMC-expressing human pituitary cell lines, DMS79 is a non-pituitary human cell line that expresses POMC and secretes measurable levels of ACTH into the culture medium. DMS79 is a human small cell lung cancer line and has been widely used as a model for ectopic ACTH producing adenomas due to its ability to express POMC and secrete ACTH from a non-pituitary origin [33–35]. Notably, several transcription factors regulate POMC expression in DMS79 cells differently from Pomc regulation in mouse AtT-20 cells [11, 35–36]. For example, Tbx19 (Tpit) is one of the key regulators of Pomc gene expression in AtT-20 cells but not in DMS79 cells [11]. While this cell line offers a platform to study human POMC regulation, it is important to note that these cells are not derived from the pituitary. As a result, POMC regulation in DMS79 cells may involve mechanisms different from those in human corticotrophs.

Transgenic Models

To study CD in vivo, several transgenic mouse or zebrafish models have been developed. These models recapitulate various aspects of CD symptoms in humans and serve as valuable tools for CD research. They are designed to induce corticotroph adenomas and/or hypercortisolemia by manipulating the functions of candidate genes associated with CD (Table 2, Figure 2).

Table 2.

Cushing’s disease/hypercortisolemia transgenic models

model	transgene	promoter	animal	phenotype	reference
Crh Tg	Chr	Mt1	mouse	hypercortisolemia	(81)
	Chr	Thy1	mouse	hypercortisolemia	(82)
	Point mutated Crh−120/+	Crh	mouse	hypercortisolemia, osteoporosis	(83)
PyLT Tg	Polyoma virus large T antigen	PE/PL	mouse	conflicting pituitary phenotype: one group: pituitary tumor, hypercortisolemia other group: testicular hyperplasia only	(56,57)
SV40LT Tg	SV40 large T antigen	AVP	mouse	pituitary and pancreas tumor, no CD phenotype	(61)
	SV40 large T antigen	Pomc	mouse	intermediate lobe tumor, CD phenotype	(63)
Rb K/O	Rb knock out		mouse	intermediate lobe tumor, CD phenotype	(67, 68, 69)
pttg Tg	pttg	Pomc	zebrafish	pituitary tumor, CD phenotype	(51)
EGFR Tg	EGFR	synthetic Pomc	mouse	pituitary tumor, CD phenotype	(78)

Models of CD with pituitary adenoma development

Large T-antigen PyLT transgenic mouse model

Polyoma large T antigen (PyLT) is responsible for viral and cellular transcriptional regulation, DNA replication, and cell cycle alteration, and has been used to immortalize various cell types. In the early 1990s, two groups (Helseth and Paquis-Fluckliner) generated PyLT transgenic mice and reported differing phenotypes with respect to CD [37, 38] (Table 2). Helseth et al. generated a transgenic mouse line using a PyLT construct containing early region and late region promoters [39], which developed pituitary adenomas around 12 months of age. These mice exhibited CD features including weight gain, adrenal medullary hyperplasia, and high levels of peripheral ACTH. Based on these phenotypes, the authors proposed the PyLT transgenic mouse line as a CD model. Conversely, Paquis-Flucklinger et al. also generated a PyLT transgenic mouse line using the same PyLT construct but did not detect pituitary adenomas [38]. Two groups also reported different testicular phenotypes; one reported normal testicular gland morphology [37], while the other reported testicular hypertrophy [38]. These contrasting phenotypes in transgenic lines established using identically designed constructs highlight the variability in outcomes. It is suggested that these phenotypic differences could be due to variations of the polyoma virus vector sequences, or differences in transgene insertion sites may have affected expression levels and tissue specificity. Although the PyLT transgenic mice are reported to be a potential CD model, we could not find literature studying CD using this transgenic line, possibly due to the above mentioned phenotypic variability.

SV40 large T antigen mouse model

The Simian virus (SV) 40 large T antigen transgenic mice have been used to develop tumors in endocrine glands such as the thyroid [40] and pancreas [41]. Two SV40 large T transgenic mouse lines have been reported in the context of CD research (Table 2). In 1987, Murphy et al. generated a transgenic mouse line using a bovine vasopressin (AVP) promoter-driven SV40 large T antigen expression [42]. AVP is known to increase ACTH secretion and enhance CRH actions on corticotrophs due to expression of vasopressin receptors [43]. AVP-SV40T mouse did not develop adenomas in the hypothalamus but did in the anterior pituitary and pancreatic beta cells. However, mice did not develop CD phenotypes, suggesting that adenomas were not derived from corticotrophs or melanotrophs [43]. In 1993, Low et al. established a transgenic mouse line expressing the SV40 large T antigen under the mouse Pomc promoter (conventional promoter, located at the 5’ of the transcription start site) [44]. Three out of 9 mice (33%) used in the study developed pituitary adenomas and developed clinical features resembling CD, including adrenal hyperplasia and increased adipose tissue. ACTH was not detected in the adenoma lysate of the SV40T transgenic mice. The authors reported that the majority of cleaved peptide products of POMC in the adenoma lysate were intermediate lobe (melanotroph)-derived peptides, such as beta endorphin, suggesting the origin of the adenoma is melanotroph, not corticotroph. [44].

Retinoblastoma (Rb) knockout mouse model

Mutations in the tumor suppressor gene Retinoblastoma (RB) are critical in the pathogenesis of many human cancers [45,46] . RB controls cell cycle progression by binding to various cell cycle regulators, including members of the E2F family. Human RB heterozygous mutations cause childhood retinoblastoma with an occurrence rate of around 90%. In 1992, Jacks et al. first presented Rb^+/− heterozygous mutant mice, which rarely developed retinoblastoma (Table 2). Instead, 25% of these mice developed pituitary adenomas [47]. Harrison et al. later reported that 90% (72 out of 80) of Rb^+/− mice developed pituitary adenomas by 8-10 months of age [48]. Expression of Pomc was detected in the pituitary adenomas obtained from all the 72 Rb^+/− mice. However, the Pomc mRNA detected in these mice was localized in melanotrophs in the intermediate lobe, not in corticotrophs in the anterior lobe. Similarly, another study reported pituitary adenomas in Rb^+/− mice, which expressed α-MSH, a melanotroph derived peptide, as shown by immunostaining [49]. These studies suggest that the pituitary adenomas found in Rb^+/− mice do not originate from corticotrophs. Therefore, it is unclear to what degree the Rb^+/− mouse model reflects human CD. In addition, despite a high frequency of occurrence of pituitary adenomas in Rb^+/− heterozygous mutant mice, a human pituitary adenoma analysis showed that the allelic defect of RB was not detected in human corticotroph adenomas [50]. This finding suggests that mutations in RB may not be a major player in human pituitary adenoma development. In summary, the above-mentioned 3 transgenic mouse models (PyLT, SV40, and Rb^+/−) developed pituitary adenomas and some clinical CD features. However, the origin of the adenomas is mainly from the intermediate lobe, and transgenes are not necessarily reflective of human CD causative genes.

Pituitary tumor transforming gene (PTTG) transgenic zebrafish model

Zebrafish possess over 26,000 protein-coding genes, and approximately 70% of human genes have a zebrafish ortholog [51–53]. Liu et al. developed a transgenic zebrafish model of CD, in which the zebrafish pituitary tumor transforming gene (pttg) is expressed under the control of the zebrafish pomc promoter [32]. PTTG is a scaffold protein that regulates mitosis, cell cycle progression, DNA repair, and apoptosis, and plays a central role in pituitary tumorigenesis [54–56]. Studies indicate that PTTG protein is expressed in up to 90% of pituitary adenomas, including corticotroph, somatotroph, lactotroph, and non-hormone producing adenomas [54]. The pttg transgenic zebrafish exhibited upregulation of the cell cycle factor cyclin E1 protein expression, in addition to elevated pomc expression levels [32]. By 3 months of age, around which zebrafish reach reproductive maturity, the pttg transgenic zebrafish presented glucocorticoid resistance, a physiological feature of an upregulated hypothalamic pituitary adrenal (HPA) axis activity and developed hypercortisolemia. Importantly, the hypercortisolemia phenotype was rescued by treating the transgenic fish with a CKD2/cyclin E inhibitor (seleciclib; R-roscovitine), indicating CKD2/cyclin E as a potential therapeutic target [32] [57]. The limitation of this model is that PTTG upregulation is not specific to human CD. A human pituitary adenoma analysis showed that PTTG gene expression is higher in other types of pituitary endocrine cells, and lower in CD [55]. Nevertheless, this model successfully developed CD features, highlights the importance of cell cycle regulation in CD, and is currently recognized as a useful transgenic model of CD.

Epidermal Growth Factor Receptor (EGFR) murine model

The epidermal growth factor receptor (EGFR) is a transmembrane receptor tyrosine kinase that activates mitogenic signaling cascades. Upregulation of EGFR expression has been reported in CD with aggressive tumor phenotypes [18,58]. To understand the role of EGFR signaling in CD, a transgenic mouse model was generated [18] . In this model, the human EGFR coding sequence is expressed in a corticotroph-specific manner using a synthetic mouse Pomc promoter with 3 repeats of TBX19/PITX1 binding sites and 11 repeats of NEUROD1 binding sites. The activity of the TBX19/PITX1 binding element in the synthetic promoter is specific to the Pomc in corticotrophs. The NEUROD1 binding element in the synthetic promoter regulates Pomc expression in a corticotroph-specific manner without inducing the expression in melanotrophs [18]. Sixty-five percent of the transgenic mice used in the study developed pituitary adenomas and clinical CD features by 6 to 7 months old. Oral administration of Gefitinib, an EGFR tyrosine kinase inhibitor, to these mice led to a reduction of adenoma size, blood ACTH levels, and blood corticosterone levels, indicating that suppressing EGFR signaling could be an effective target in CD treatment [24,18]. Expression of E2f1, a cell cycle regulator involved in G1/S transition, was elevated in the pituitary adenomas of the Pomc-EGFR transgenic mice, suggesting that the E2F1 could be downstream of EGFR signaling in this model. In humans, E2F1 is highly expressed in normal corticotrophs by immunostaining of autopsied pituitaries, and nuclear translocation of E2F1 is observed in human CD [59]. These observations suggest that EGFR transgenic mice may recapitulate human CD with activation of E2F1. These studies support the idea that the EGFR-E2F1 axis is a promising therapeutic target of CD. A limitation of this model is that EGFR overexpression is not a common feature of all human corticotroph adenomas. EGFR is expressed in a subgroup of corticotroph adenomas in humans, especially in aggressive CD. Hayashi et al. immunostained 60 human corticotroph adenomas and reported 9 out of 60 (15%) of the adenomas express at least low levels of EGFR protein [60]. Although the Pomc-EGFR transgenic mouse line represents only a fraction of CD cases, this model suggests that EGFR is one of the drivers of corticotroph adenoma development, and investigation using this mouse model is relevant to understanding the genetic and molecular mechanisms of CD pathogenesis.

Models of hypercortisolemia

In addition to the corticotroph adenoma model animals described above, the research community has also developed mouse models that mimic hypercortisolism. While these models do not develop corticotroph adenomas, they enable the investigation of the physiological sequelae of hypercortisolemia. We briefly summarize these models below.

Corticotropin Releasing Hormone (CRH) transgenic mouse models.

Three mouse models have been developed to mimic hypercortisolism characteristic of Cushing’s syndrome [61–63] (Table 2). These models do not develop corticotroph adenomas; instead, their phenotypes are caused by elevated levels of corticotropin-releasing hormone (CRH).

The first model, developed by Stenzel-Poore et al. in 1992, is a Crh transgenic mouse model. The transgene consists of the rat Crh coding sequence (1700 bp), driven by the mouse metallothionine 1 (Mt1) promoter [61]. The Crh transgene was expressed in the pituitary and hypothalamus, with low levels of expression detected in the testicular and adrenal glands. The Mt1-Crh transgenic mice manifest anxiety, obesity, osteoporosis, and muscle weakness, with elevated serum ACTH and corticosterone levels. These phenotypes recapitulate CD symptoms in humans.

The second Crh transgenic mouse model was generated by Groenink et al. in 2002 [62]. In this model, Crh expression is driven by the Thy1 promoter (Thy1-Crh), which constitutively expresses Crh in all neurons of neonates and adults. The Thy1-Crh mice exhibit features of an activated HPA axis, including elevated levels of serum ACTH and corticosterone. Due to the non-physiological regulation of the Thy1-Crh transgene, these mice also fail to suppress corticosterone levels upon dexamethasone administration. This loss of negative feedback mimics a clinical feature of CD, and this model has been instrumental in assessing stress reactivity under conditions of hypercortisolemia.

The third model was generated by N-ethyl-N-nitrosourea (ENU)-induced random mutagenesis in 2014 [63]. In this mutant mouse line, a base substitution in the Crh promoter (T replaced by C at the −120 position from the transcription start site, Crh^−120/+) caused a twofold increase in Crh promoter activity (gain-of-function), leading to hypersecretion of CRH. Elevated levels of CRH peptide subsequently increased ACTH secretion, resulting in elevated peripheral corticosteroid levels. The mutant mice exhibited typical phenotypes of Cushing’s syndrome, including loss of circadian rhythm, elevated plasma corticosterone levels without developing pituitary adenomas or hyperplasia [63]. Interestingly, Crh^−120/+ mice showed their low parathyroid hormone (PTH) levels, hypercalciuria, and low bone density, which distinguish them from the other models described above. Consequently, this Crh promoter mutant model is valuable for evaluating treatments for corticosteroid-induced osteoporosis.

Emerging human cell-derived models

While the models discussed above are valuable, the extent to which each model accurately recapitulates human CD remains to be determined. Therefore, there has been a demand for models utilizing human cells to develop more accurate CD models (Table 3, Figure 2).

Table 3.

Comparison of pituitary organoid development methods

culture methods	2D	2D transfer to 3D	de-novo 3D
In vivo pituitary structure	no	yes	yes
Pituitary with hypothalamus	no	no	yes
>1 year cell survival	N/A	N/A	yes

Sustainable primary culture of ACTH producing corticotroph cells

The first approach using human cells for CD research involves primary cultures of human corticotroph adenomas. Usually, traditional 2-dimensional (2D) harvested primary culture systems allow cells to survive only 48-72 hours [65]. Recently, in vitro 3D explant culture methods have been developed for various types of cancers, including colon [66], pancreas [67], and brain cancers [68]. In the context of corticotroph adenomas, Zhang et al. developed a CD 3D primary culture model using Matrigel as a scaffold matrix, supporting cultured explants for up to 18 weeks [65]. This CD 3D primary culture system represents a novel approach for studying human corticotroph adenoma cells and may be useful for assessing drug efficacy in human adenomas. The limitations of this method are that ACTH secretion diminishes after 20 weeks of culture, and the availability of materials (i.e., surgical specimens, which are generally small in size) is limited.

Human induced pluripotent stem cell (hiPSC)-derived corticotroph organoids

Recently, human pluripotent stem cells (hPSCs) have emerged as a potential tool for studying CD by deriving pituitary hormone-producing cells from hPSCs in vitro. hPSCs can differentiate into any somatic cell type and are grouped into 2 types of PSCs; human embryonic stem cells (hESCs) and induced pluripotent stem cells (hiPSCs). To date, several induction methods for pituitary hormone-producing cells from hPSCs have been reported. These novel stem cell technologies, combined with the advent of genome engineering technologies, such as the CRISPR/Cas9 system, has provided an opportunity to generate genetic models of CD. There are 3 methods to direct hPSCs into pituitary hormone-producing cells: 2D, 2D transfer to 3D, and de-novo 3D cultures. Comparisons of the methods are summarized in Table 3.

In 2013, Dincer et al. reported the first 2D pituitary derivation method from hESCs [70]. Previously, they have developed an induction method for neural ectoderm cells from hESCs with high efficiency by inhibiting BMP and TGFβ signaling (dual SMAD inhibitions) [69]. They modified that method and established a further induction method into cranial placodal tissues by removing BMP signal inhibition. Furthermore, activation of sonic hedgehog signaling in the cells resulted in induction into pituitary hormone-producing cells [69]. The induced ACTH- or GH-producing cells, transplanted into pituitary insufficiency mouse models, could survive in the host (the mouse CNS) and exhibit hormone secretion capabilities (ACTH or GH) [70]. Since this protocol allows for producing somatotrophs or corticotrophs, it provides a novel platform for studying pituitary physiology and disorders. However, the 2D culture techniques cannot represent pituitary tissue architecture, and maintaining the induced hormone-producing cells in long-term culture is still challenging.

To address the limitations of the 2D culture system, researchers have developed a method where hormone-producing cells, induced in 2D culture, are maintained within a 3D organoid culture system using Matrigel as a scaffold (2D transfer to 3D method) [71]. Mallick et al. successfully generated hiPSC-derived corticotroph organoids using this method to study the effects of glucocorticoid receptor (GR) in corticotroph adenomas [72]. Using CRISPR/Cas9 gene editing, they generated genetically engineered iPSCs expressing mutant USP48 or USP8, genes frequently mutated in CD. These engineered iPSCs were subsequently induced into corticotroph organoids to establish an in vitro corticotroph adenoma model.

An advantage of this method is that organoids can last longer than the 2D system, making it more suitable for drug response testing or drug screening. However, organoids from this method lacks the formation of surrounding pituitary tissues (i.e. hypothalamus), therefore, it cannot maintain the cells long enough to observe long-term changes. Although there is no literature describing the exact duration of maintenance of the cells, 2 studies using these methods evaluated corticotroph organoids at 30 days post-differentiation [71, 72].

Finally, Suga et al. have established a de novo 3D pituitary organoid induction method to recapitulate the complex interactions involved in pituitary development [73]. This method enables the generation of both anterior pituitary cells (corticotrophs or somatotrophs) and surrounding hypothalamic cells within the same organoid During embryonic development, interactions between the oral ectoderm, the precursor of the anterior lobe of the pituitary, and the prosencephalic neuroectoderm, the precursor of the hypothalamus, are essential for proper pituitary development [74]. The de novo 3D method mimics this interaction by simultaneously inducing the oral ectoderm and hypothalamus while facilitating cell differentiation into the pituitary lineage (anterior pituitary self-organization), ultimately leading to the formation of corticotrophs or somatotrophs.

De-novo 3D model can be used to study various pituitary–hypothalamic conditions. For example, using this method, a congenital hypopituitarism model was generated from patient-derived iPSCs harboring an Orthodentible Homoeobox2 (OTX2) mutation, a key gene involved in pituitary development [75]. It has also been applied to study the pathogenesis of acquired autoimmune hypophysitis [76]. To the best of our knowledge, there have been no reports of its application to corticotroph adenomas. However, by introducing genetic mutations associated with human pituitary adenomas in hiPSCs, this 3D system holds potential as a valuable tool for studying tumorigenesis. Another advantage of hiPSC-derived organoids generated using de novo 3D methods is their extended maintenance period, up to 500 days [77], allowing for long-term monitoring of organoid phenotypes. However, a major limitation of these hiPSC-based induction methods is the extended culture period required to obtain pituitary hormone-producing cells, leading to increased researcher workload and higher costs (Table 3, Figure 2).

Conclusion

We summarized 11 various models of CD within detailed discussion of advantages and limitations. Much remains to be discovered about the pathogenesis of CD, a rare and challenging neuroendocrine disease. A deeper understanding of CD pathophysiology can lay the foundation to develop novel therapeutic approaches to CD, highlighting the importance of developing more relevant disease models.